Hi Ted,
I am running a simulation already for 1 year and everything was quite ok. Now I decided to increase the running time from tstop=4e4(ms) to tstop=1.6e5(ms) and I already realized that at time t=1.53e5(ms) everything suddenly closes, I mean the whole GUI of NEURON and even the console. What could be a reason for that?
Thanks a lot for your help in advance
Oscar Javier
NEURON GUI suddenly closes; why..?
-
- Site Admin
- Posts: 6299
- Joined: Wed May 18, 2005 4:50 pm
- Location: Yale University School of Medicine
- Contact:
Re: NEURON GUI suddenly closes; why..?
Very strange. To identify the cause, I'll have to be able to reproduce the symptom. Can you zip up the mod, hoc, and ses files that are necessary to run the simulation, and send them to me
ted dot carnevale at yale dot edu
?
Also, what version of NEURON are you using?
ted dot carnevale at yale dot edu
?
Also, what version of NEURON are you using?
Re: NEURON GUI suddenly closes; why..?
Hi Again Ted,
I have been looking at my simulation and the problem seem to be related with the "weight" of external drive in the form of spikes trains. To deliver the spikes I am creating one NetStim object such that
tstop=2e5ms-2.5e5
.
.
.
interval=90ms
number=tstop/interval
noise=1
This connected to a Expsyn with e=0V tau=2ms and w=3.3e-4 uS, this last parameter appears to be responsible for the crashing of the whole simulation and even the sudden closing of the GUI and console after 1.53e5ms. The strange thing is that by reducing w from 3.3e-4 uS to 1.6e-4uS the simulation runs at least until 2.5e5ms i.e. about the double of the original time.
I think it could be one problem of instability during the computation of the exponential solution of the derivative in the Expsyn object but I can't really see a clear relation between the problem and the object,
The second possibility is that the problem is related to the definition of my gaba synapses and although I haven't found suspicious stuff maybe something in the state discontinuity of the NET_RECEIVE block can be producing divergences in during the computation, here a copy of the used GABAA receptor definition
One more question: Is it possible to get a log file for each run of the simulation?
Thanks again for any clue
Oscar Javier
I have been looking at my simulation and the problem seem to be related with the "weight" of external drive in the form of spikes trains. To deliver the spikes I am creating one NetStim object such that
tstop=2e5ms-2.5e5
.
.
.
interval=90ms
number=tstop/interval
noise=1
This connected to a Expsyn with e=0V tau=2ms and w=3.3e-4 uS, this last parameter appears to be responsible for the crashing of the whole simulation and even the sudden closing of the GUI and console after 1.53e5ms. The strange thing is that by reducing w from 3.3e-4 uS to 1.6e-4uS the simulation runs at least until 2.5e5ms i.e. about the double of the original time.
I think it could be one problem of instability during the computation of the exponential solution of the derivative in the Expsyn object but I can't really see a clear relation between the problem and the object,
The second possibility is that the problem is related to the definition of my gaba synapses and although I haven't found suspicious stuff maybe something in the state discontinuity of the NET_RECEIVE block can be producing divergences in during the computation, here a copy of the used GABAA receptor definition
Code: Select all
TITLE first_GABA_recept_model (first order kinetics )
COMMENT
This program just explore basics charactieristics of GABAergic synapses
in the frame of a simple integrate and fire model. Modified from original of R. van Elburg and based on
Bush PC, Prince DA, Miller KD (1999),J Neurophysiol 82:1748-58
This is mod-file describes a point process with first order kinetics, that can act as several
synapses described by first order kinetics.
Author:
Taken from model for pyramidal cells in J. Tegner, A. Compte,
X.J. Wang, J. Neurosci. 22(20): 9053-9062, 2002
X.J. Wang, J. Neurosci. 21(19):9587-9603
Modifications:
ID Date Authors Email Description
M_001
ENDCOMMENT
NEURON {
POINT_PROCESS Gaba_syn
RANGE e, i, g
RANGE tau_d, frac_rec
RANGE area_cell
NONSPECIFIC_CURRENT i
}
UNITS {
(nA) = (nanoamp)
(mV) = (millivolt)
(uS) = (microsiemens)
(um) = (micron)
}
PARAMETER {
: e = -90 mV for inhibitory synapses,
: 0 mV for excitatory
:e = :-90
e=-80 (mV): value of boergers and koppel's paper
: tau_d decay time, frac_rec usage fraction of receptors
tau_d = 10 (ms) < 1e-9, 1e9 > :
frac_rec = 0.9 (1) <0,1>
: maximum coductance
g=1e-5 (uS/um2) : this must be multiplied by the area to give actual walue
: since the conductance density is gie=5-7mS/cm2 gii=15-20mS/cm2
: Jensen et al. NeuroImage 26 (2005) 347-355
:cell surface area
area_cell= 1 (um2)
}
ASSIGNED {
v (mV)
i (nA)
g_eff (uS)
}
STATE {
s
}
INITIAL {
s=0 :
g_eff=g*area_cell
}
BREAKPOINT {
SOLVE state METHOD cnexp
i =g_eff*s*(v - e) :
}
DERIVATIVE state {
s' = -s/tau_d
}
NET_RECEIVE(weight,s_tp,tp(ms)) {:tp time of previous spike
: Calculate current value single synapse state variable at t-epsilon
UNITSOFF
:printf("%g %g\t",t,s_tp)
s_tp =s_tp*weight*exp(-(t-tp)/tau_d)
:printf("%g\t",s_tp)
UNITSON
: To make sure that we add the same amount to the exact single synapse state variable and the summed
: state variable we should first update the summed variable and then the single synapse state variable
state_discontinuity(s, s + frac_rec*weight*(1-s_tp))
s_tp = s_tp + frac_rec*(1-s_tp)
:printf("%g %g %g %g\n",weight,s_tp,s,frac_rec)
tp=t
}
Thanks again for any clue
Oscar Javier